Re: [PATCH v3 1/2] net/i40e: new API to add VF MAC address from PF

Hi,

-----Original Message-----
From: Yigit, Ferruh
Sent: Friday, August 18, 2017 12:43 PM
To: Stephen Hemminger <stephen@networkplumber.org>; Lu, Wenzhuo
[off-list ref]
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH v3 1/2] net/i40e: new API to add VF MAC
address from PF

On 8/18/2017 1:32 AM, Stephen Hemminger wrote:

quoted

On Fri, 18 Aug 2017 02:33:42 +0800
Wenzhuo Lu [off-list ref] wrote:

quoted

Currently, rte_eth_dev_mac_addr_add is used by a testpmd CLI to add a
MAC address for VF. But the parameter 'pool' of this API means the
VMDq pool, not VF.
So, it's wrong to use it to add the VF MAC address.
This patch provides a new API that can be used to add VF MAC address
on i40e.

Signed-off-by: Wenzhuo Lu <redacted>

What do other drivers do?

This is adding VF MAC through PF, not all PMDs support this.

quoted

Sorry, a driver specific API is (almost) always the wrong solution.

Currently there are set of device specific APIs [1].

The motivation was to be able to expose NIC specific features, abstraction
layer should not prevent using HW features; while keeping abstraction clean
for common majority, not having APIs valid only for a single NIC.

But with number of the device specific APIs increasing, it may be time to
bring postponed discussion to life on how to manage them.

Recent ioclt or staging dev_ops was a good tart:
http://dpdk.org/ml/archives/dev/2017-August/thread.html#72423


[1]
Physical PMDs with device specific API: i40e, ixgbe, bnxt Virtual PMDs with
device specific API: ring, vhost, bonding, xenvirt

This patch follows the existing bnxt example to implement the same function on i40e.

Totally agree the private API is not a good idea. To my opinion, as many same functions are implemented on ixgbe, i40e, bnxt, it's them time to make them public. Let me send some RFC later.